The lists
| List | Category | Domains | Source |
|---|---|---|---|
| Adult | adult | 61,159 | extensions/porn/sinfonietta |
| Violence | violence | 23 | extensions/porn/sinfonietta-snuff |
| Gambling | gambling | 2,690 | extensions/gambling/sinfonietta |
The lists come from the StevenBlack/hosts project, published under the MIT licence.
How matching works
A destination matches if its hostname, or any parent domain of it, is on a list. A link to shop.example-casino.com matches if example-casino.com is listed.
A match is a confirmed verdict: the link is marked Not safe and the detail panel shows a Confirmed tag. It takes priority over everything else, including a page that returned an error.
What the lists don’t cover
- Pharma and payday-loan content has no blocklist. Those categories rely on keywords and AI analysis.
- The lists are domain-level. They cannot say that one page on an otherwise ordinary site is bad.
- New domains and large brands may be missing. See accuracy and limits.
Where the lists come from and how they are prepared
StevenBlack/hosts is a long-running open-source project that merges many community-maintained lists into hosts files, with optional extensions for categories such as adult content and gambling. Link Signals downloads three of those extension files. A hosts file is a plain list of lines such as 0.0.0.0 example-casino.com, written for blocking at the network level. For matching, each line is reduced to just the domain name, converted to lower case, and de-duplicated, and anything that is not a real dotted domain (such as localhost) is dropped. What is left is a flat list of domains for each category.
The lists are refreshed by hand, not live. The caption of the table at the top of this page states the date of the last refresh, and the counts show how large each list was then. A domain added to the upstream project after that date is not matched until the next refresh.
What a match looks like in a report
A blocklisted destination is Not safe and carries a Confirmed tag. If the page carrying the link is not itself about that category, the reason is written as an out-of-context link. If it is (a gambling-review site linking to a casino, say), the reason is simply that the domain is on the gambling blocklist, because that link is expected.
Domain is on the gambling blocklist
/reviews/best-online-casinos · Anchor text “Play Royal Slots”The same applies to a link that goes to a subdomain. If example-casino.com is listed, then www.example-casino.com and promo.example-casino.com match too.
Why a match wins over everything else
A blocklist match is checked before the page is judged and even before an error status is considered. A listed domain whose server is down, or which answers with a 403, is still Not safe, because the reason to distrust it has nothing to do with how the page responds. A blocklisted domain is also not looked up in the domain registry, since it is already confirmed. The one thing that comes before a match is an opt-out: a site whose owner has asked not to be requested is never fetched, but if it is on a list it is still reported from the list alone.
How the lists fit with the other checks
| Check | What it catches | How strong the evidence is |
|---|---|---|
| Blocklist | Known adult, gambling and violent-content domains. | Strongest: Confirmed. |
| AI analysis | Pages that are unsafe or unrelated to where they are linked from, including categories with no list. | Medium: the detail panel says the model judged it. |
| Keywords | Risky words on the destination page, or in its domain name, in categories such as adult, gambling, pharma and payday loans. | Weakest: always Needs review, never Not safe on their own. |
| Domain registration | Domains that are unregistered, expired, very new, or parked. | Strong for the fact, but it is a risk signal, not a content judgement. |
A worked example
A garden centre’s old news post links to www.deals-for-you.example. The domain is on none of the lists, so there is no Confirmed match. The page loads and advertises an online pharmacy and tablets with no prescription needed, which hit the pharma keywords. The detail panel shows “Matched:” followed by those words, and the verdict is Needs review. If AI analysis is on, the model also sees that a garden news post has no reason to link to a pharmacy and records its reason. Either way, no blocklist was involved, because there is no pharma list. This is why the pharma and payday-loan categories need a person to look.
Questions
Why do you use these lists and not a paid threat feed?
They are public, openly licensed and well maintained, and they cover the categories where a confirmed domain match is most useful. The trade-off is that they focus on adult and gambling content, which is why other categories rely on keywords and AI analysis.
My site was flagged by mistake. Can I be removed?
Link Signals only reads the lists. If you think a domain is listed in error, ask the upstream project, StevenBlack/hosts, which maintains them. On your own report you can ignore a finding so that monitoring stops alerting on it.
Does a match mean the site is dangerous to visit?
Not necessarily. Adult and gambling sites are often legal. The match means the content is probably unsuitable for the kind of site that links to it, especially if the link was added without anyone meaning to.
Are malware and phishing sites on the lists?
No. These three lists cover adult, gambling and violent content. Malware and phishing are judged by AI analysis from the page itself, which is less certain than a list match.
How often are the lists refreshed?
At intervals, by hand. The date of the latest refresh is in the caption of the table at the top of this page.
See which of your outbound links match the lists.
